\u3053\u308c\u3060\u3051\u306f\uff08\u30dd\u30a4\u30f3\u30c8\uff09 \u672c\u8a18\u4e8b\u3067\u8a71\u3059\u5185\u5bb9\u306e\u4e2d\u3067\u3001\u91cd\u8981\u306a\u90e8\u5206\u3092\u30d4\u30c3\u30af\u30a2\u30c3\u30d7\u3057\u3066\u7b87\u6761\u66f8\u304d\u306b\u3057\u307e\u3057\u305f\u3002 \u9806\u4f1d\u64ad\u578b\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\uff08FNN\uff09 \u9806\u4f1d\u64ad\u578b\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\uff08Feedforward Ne ... <\/p><\/div><\/dd><\/dl><\/div><\/a><\/div>\n<\/div><\/figure>\n\n\n\n
MNIST\u3067\u6642\u7cfb\u5217\u51e6\u7406\u3092\u884c\u3046\u5834\u5408\u306f\u3001\u4e0a\u3067\u8aac\u660e\u3057\u305f\u3088\u3046\u306b\u30d9\u30af\u30c8\u30eb\u306b\u5206\u3051\u3001\u6642\u7cfb\u5217\u5206\u6790\u3092\u884c\u3044\u307e\u3059\u3002<\/p>\n\n\n\n
\u203b\u300c\u3044\u3061\u3001\u306b\u3001\u3055\u3093\u3001\uff0e\uff0e\uff0e\u300d\u3068\u3044\u3046\u81ea\u7136\u8a00\u8a9e\u3092\u30c8\u30fc\u30af\u30f3\u5316\u53ca\u3073\u30d9\u30af\u30c8\u30eb\u5316\u3059\u308b\u3068\u3001\u5404\u6570\u5b57\u306b1\u3064\u306e\u30d9\u30af\u30c8\u30eb\u304c\u5272\u308a\u5f53\u3066\u3089\u308c\u308b\u3060\u3051\u306b\u306a\u308b\u3060\u3051\u306a\u306e\u3067\u30011\u3064\u306e\u6570\u5b57\u309228\u6642\u523b\u5206\u306e\u30d9\u30af\u30c8\u30eb\u306e\u6642\u7cfb\u5217\u3067\u8868\u3059\u306b\u306f\u7121\u7406\u304c\u3042\u308a\u307e\u3059\u304c\u3001\u4eca\u56de\u306f\u3001\u30d9\u30af\u30c8\u30eb\u5316\u3057\u305f\u7d50\u679c\u300128\u6642\u523b\u5206\u306e\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u306b\u306a\u3063\u305f\u3068\u4eee\u5b9a\u3057\u307e\u3059\u3002<\/p>\n\n\n\n
\u4eca\u56de\u306f\u3001keras\u3092\u4f7f\u7528\u3057\u3066\u3001\u30e2\u30c7\u30eb\u3092\u69cb\u7bc9\u3057\u3066\u3044\u304d\u307e\u3059\u3002<\/p>\n\n\n\n
\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u4ee5\u4e0b\u306b\u793a\u3057\u307e\u3059\u3002<\/p>\n\n\n\n
from tensorflow.keras.models import Sequential\nfrom tensorflow.keras import layers\nfrom tensorflow.keras.optimizers import RMSprop\nfrom tensorflow.keras.datasets import mnist\nfrom tensorflow.keras.utils import to_categorical\n\n# \u5b66\u7fd2\u30c7\u30fc\u30bf\u3092\u30ed\u30fc\u30c9\n(x_train, y_train), (x_test, y_test) = mnist.load_data()\n# \u6b63\u898f\u5316\u3059\u308b\nx_train = x_train \/ 255\nx_test = x_test \/ 255\n# onehot\u30d9\u30af\u30c8\u30eb\u8868\u73fe\u306b\u3059\u308b\ny_train = to_categorical(y_train)\ny_test = to_categorical(y_test)\n\n# \u30e2\u30c7\u30eb\u3092\u5b9a\u7fa9\nmodel = Sequential()\nmodel.add(layers.Conv1D(32, 7, activation='relu', input_shape=(28, 28)))\nmodel.add(layers.MaxPool1D(3))\nmodel.add(layers.Conv1D(32, 5, activation='relu'))\nmodel.add(layers.GlobalAveragePooling1D())\nmodel.add(layers.Dense(10, activation='softmax'))\n\nmodel.summary()\n\nmodel.compile(optimizer=RMSprop(lr=1e-4),\n loss='categorical_crossentropy',\n metrics=['acc'])\n\n# \u30e2\u30c7\u30eb\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3092\u5b9f\u65bd\nhistory = model.fit(x_train, y_train, epochs=100, batch_size=128, validation_split=0.2)<\/code><\/pre><\/div>\n\n\n\nsummary\u3067\u8868\u793a\u3055\u308c\u308b\u3082\u306e\u3092\u793a\u3057\u307e\u3059\u3002conv1d_1\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u6570\u306e\u6c42\u3081\u65b9\u3092\u78ba\u8a8d\u3057\u3066\u304a\u304d\u307e\u3059\u3002\u30a6\u30a3\u30f3\u30c9\u30a6\u30b5\u30a4\u30ba7\u306e1\u6b21\u5143\u7573\u307f\u8fbc\u307f\u30d5\u30a3\u30eb\u30bf\u3092\u30c1\u30e3\u30f3\u30cd\u30eb\u657028\u306e\u30c7\u30fc\u30bf\uff08\u30d9\u30af\u30c8\u30eb\u306e\u6210\u5206\u6570\uff09\u306b\u9069\u7528\u3059\u308b\u3053\u3068\u3092\u8003\u3048\u3066\u304a\u308a\u3001\u305d\u308c\u309232\u7a2e\u985e\u9069\u7528\u3059\u308b\u306e\u3067\u3001<\/p>\n\n\n\n
\\(
\\begin{eqnarray}
\\begin{array}{rcl}
&&\\mbox{window_size}\u00d7\\mbox{input_channels}\u00d7\\mbox{output_channels}+\\mbox{num_bias}\\\\
&=&7\u00d728\u00d732+32\\\\
&=&6304
\\end{array}
\\end{eqnarray}
\\)<\/p>\n\n\n\n
\u3068\u306a\u308a\u307e\u3059\u30022\u3064\u76ee\u306econv1d_2\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u6570\u306f\u3001\\(5\u00d732\u00d732+32=5152\\)\u3067\u3059\u3002<\/p>\n\n\n\n
Model: "sequential_1"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\nconv1d_1 (Conv1D) (None, 22, 32) 6304 \n_________________________________________________________________\nmax_pooling1d_1 (MaxPooling1 (None, 7, 32) 0 \n_________________________________________________________________\nconv1d_2 (Conv1D) (None, 3, 32) 5152 \n_________________________________________________________________\nglobal_average_pooling1d_1 ( (None, 32) 0 \n_________________________________________________________________\ndense_1 (Dense) (None, 10) 330 \n=================================================================\nTotal params: 11,786\nTrainable params: 11,786\nNon-trainable params: 0\n_________________________________________________________________<\/code><\/pre><\/div>\n\n\n\n\u5b66\u7fd2\u6700\u4e2d\u306b\u8868\u793a\u3055\u308c\u308b\u30ed\u30b0\u3092\u6700\u521d\u3068\u6700\u5f8c\u306e\u307f\u793a\u3057\u307e\u3059\u3002\u6ce8\u76ee\u3057\u3066\u3044\u305f\u3060\u304d\u305f\u3044\u306e\u306f\u3001\u30e2\u30c7\u30eb\u306e\u51e6\u7406\u306e\u8efd\u3055\u3068\u5b66\u7fd2\u306e\u901f\u3055\u3067\u3059\u30021\u30a8\u30dd\u30c3\u30af\u304c3\u79d2\u3067\u7d42\u308f\u3063\u3066\u3044\u3066\u3001\u307e\u305f\u4ee5\u4e0b\u3067\u306f\u7701\u7565\u3057\u3066\u3044\u307e\u3059\u304c\u30015\u30a8\u30dd\u30c3\u30af\u76ee\u3001\u3059\u306a\u308f\u3061\u5b66\u7fd2\u958b\u59cb\u304b\u308915\u79d2\u5f8c\u306b\u306f\u691c\u8a3c\u30c7\u30fc\u30bf\u3067\u306e\u6b63\u89e3\u7387\u304c91%\u306b\u306a\u308a\u307e\u3057\u305f\u3002\u5b66\u7fd2\u306b\u8981\u3057\u305f\u6642\u9593\u306f\u7d045\u5206\u3067\u3059\u3002(\u3053\u3053\u3067\u306fRMSProp\u3092\u4f7f\u7528\u3057\u307e\u3057\u305f\u304c\u3001Adamax\u3092\u4f7f\u7528\u3057\u3066\u6700\u9069\u5316\u3057\u305f\u30892\u30a8\u30dd\u30c3\u30af\u76ee\u306790%\u4ee5\u4e0a\u306e\u6b63\u89e3\u7387\u3092\u9054\u6210\u3057\u307e\u3057\u305f\u3002)<\/p>\n\n\n\n
Epoch 1\/100\n375\/375 [==============================] - 3s 8ms\/step - loss: 1.9473 - acc: 0.4046 - val_loss: 1.4880 - val_acc: 0.6587\nEpoch 2\/100\n375\/375 [==============================] - 3s 7ms\/step - loss: 1.1329 - acc: 0.7218 - val_loss: 0.8373 - val_acc: 0.7905\nEpoch 3\/100\n375\/375 [==============================] - 3s 7ms\/step - loss: 0.7130 - acc: 0.8121 - val_loss: 0.5769 - val_acc: 0.8497\n\u30fb\n\u30fb\n\u30fb\nEpoch 98\/100\n375\/375 [==============================] - 3s 7ms\/step - loss: 0.0522 - acc: 0.9846 - val_loss: 0.0661 - val_acc: 0.9817\nEpoch 99\/100\n375\/375 [==============================] - 3s 7ms\/step - loss: 0.0519 - acc: 0.9847 - val_loss: 0.0662 - val_acc: 0.9809\nEpoch 100\/100\n375\/375 [==============================] - 3s 7ms\/step - loss: 0.0516 - acc: 0.9845 - val_loss: 0.0659 - val_acc: 0.9817<\/code><\/pre><\/div>\n\n\n\n\u7d50\u679c\u306e\u30b0\u30e9\u30d5\u3092\u4ee5\u4e0b\u306e\u30d7\u30ed\u30b0\u30e9\u30e0\u3067\u8868\u793a\u3057\u3066\u307f\u307e\u3059\u3002<\/p>\n\n\n\n
import matplotlib.pyplot as plt\nimport numpy as np\n\nacc = history.history['acc']\nloss = history.history['loss']\nval_acc = history.history['val_acc']\nval_loss = history.history['val_loss']\n\n# x\u8ef8\u3068\u3057\u3066epochs\u3092\u7528\u610f\nepochs = np.arange(1, len(acc) + 1, 1)\n\nplt.figure(figsize=(20, 5))\nplt.subplot(1, 2, 1)\nplt.plot(epochs, acc, label='training_acc')\nplt.plot(epochs, val_acc, label='validation_acc')\nplt.xlabel('epochs')\nplt.ylabel('accuracy [%]')\nplt.grid()\nplt.legend()\n\nplt.subplot(1, 2, 2)\nplt.plot(epochs, loss, label='training_loss')\nplt.plot(epochs, val_loss, label='validation_loss')\nplt.xlabel('epochs')\nplt.ylabel('loss')\nplt.grid()\nplt.legend()<\/code><\/pre><\/div>\n\n\n\n
![\"\"](\"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/08\/1dCNN_results-3.png\")
<\/figure><\/div>\n\n\n\u3053\u308c\u3060\u3051\u898b\u305b\u3089\u308c\u3066\u3082\u6bd4\u8f03\u5bfe\u8c61\u304c\u306a\u3044\u30681\u6b21\u5143CNN\u306e\u51c4\u3055\u304c\u5206\u304b\u3089\u306a\u3044\u3068\u601d\u3046\u306e\u3067\u3001\u8a08\u7b97\u30b3\u30b9\u30c8\u304c\u8efd\u304f\u3066\u6027\u80fd\u304c\u826f\u3044GRU\u3068\u6bd4\u8f03\u3057\u3066\u307f\u305f\u3044\u3068\u601d\u3044\u307e\u3059\u3002<\/p>\n\n\n\n
GRU\u306e\u5834\u5408<\/h3>\n\n\n\n
\u4e0a\u8a18\u306e\u30d7\u30ed\u30b0\u30e9\u30e0\u5185\u306b\u304a\u3044\u3066\u3001model\u306e\u90e8\u5206\u3092\u4ee5\u4e0b\u3067\u5b9a\u7fa9\u3059\u308bmodel_1\u306b\u66f8\u304d\u63db\u3048\u307e\u3059\u3002<\/p>\n\n\n\n
model_1 = Sequential()\nmodel_1.add(layers.GRU(32, activation='relu', return_sequences=True, input_shape=(28, 28)))\nmodel_1.add(layers.GRU(32, activation='relu'))\nmodel_1.add(layers.Dense(10, activation='softmax'))<\/code><\/pre><\/div>\n\n\n\n\u3053\u306e\u3068\u304d\u3001summary\u306f\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u306a\u308a\u307e\u3059\u3002\u5148\u306b\u793a\u3057\u305f1\u6b21\u5143\u7573\u307f\u8fbc\u307f\u306e\u6642\u306esummary\u3068\u6bd4\u8f03\u3057\u3066\u3044\u305f\u3060\u304f\u3068\u3001\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u6570\u304c\u307b\u307c\u540c\u3058\u3067\u3042\u308b\u3053\u3068\u304c\u5206\u304b\u308b\u3068\u601d\u3044\u307e\u3059\u3002\u7573\u307f\u8fbc\u307f\u30d5\u30a3\u30eb\u30bf\u306e\u5834\u5408\u3001\u4ee5\u524d\u306e\u8a18\u4e8b\u3067\u8aac\u660e\u3057\u3066\u3044\u307e\u3059\u304c\u3001\u91cd\u307f\u5171\u6709\u304c\u884c\u308f\u308c\u308b\u305f\u3081\u3001\u5b9f\u969b\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u6570\u3088\u308a\u3082\u306f\u308b\u304b\u306b\u5927\u304d\u306a\u8868\u73fe\u80fd\u529b\u304c\u3042\u308b\u3053\u3068\u304c\u4e88\u60f3\u3055\u308c\u308b\u305f\u3081\u3001\u30d1\u30e9\u30e1\u30fc\u30bf\u6570\u306e\u307f\u3067\u6bd4\u8f03\u3059\u308b\u306e\u304c\u826f\u3044\u3068\u306f\u8a00\u3048\u307e\u305b\u3093\u304c\u3001\u4e00\u3064\u306e\u6307\u6a19\u3068\u3057\u3066\u6bd4\u8f03\u3057\u3066\u307f\u307e\u3059\u3002<\/p>\n\n\n\n
Model: "sequential_2"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ngru (GRU) (None, 28, 32) 5952 \n_________________________________________________________________\ngru_1 (GRU) (None, 32) 6336 \n_________________________________________________________________\ndense_2 (Dense) (None, 10) 330 \n=================================================================\nTotal params: 12,618\nTrainable params: 12,618\nNon-trainable params: 0\n_________________________________________________________________<\/code><\/pre><\/div>\n\n\n\n\u5b66\u7fd2\u30ed\u30b0\u306e\u6700\u521d\u3068\u6700\u5f8c\u3092\u793a\u3057\u307e\u3059\u30021\u30a8\u30dd\u30c3\u30af\u7d42\u4e86\u3059\u308b\u306e\u306b9\u79d2\u307b\u3069\u3067\u3001\u691c\u8a3c\u30c7\u30fc\u30bf\u306e\u6b63\u89e3\u7387\u304c90%\u3092\u8d85\u3059\u306e\u306f\u300112\u30a8\u30dd\u30c3\u30af\u76ee\u3067\u3059\u3002<\/p>\n\n\n\n
Epoch 1\/100\n375\/375 [==============================] - 9s 24ms\/step - loss: 2.2791 - acc: 0.1755 - val_loss: 2.2371 - val_acc: 0.2853\nEpoch 2\/100\n375\/375 [==============================] - 9s 23ms\/step - loss: 2.0579 - acc: 0.3122 - val_loss: 1.7070 - val_acc: 0.4748\nEpoch 3\/100\n375\/375 [==============================] - 9s 23ms\/step - loss: 1.3151 - acc: 0.5895 - val_loss: 0.9624 - val_acc: 0.6867\n\u30fb\n\u30fb\n\u30fb\nEpoch 98\/100\n375\/375 [==============================] - 9s 23ms\/step - loss: 0.0651 - acc: 0.9792 - val_loss: 0.0834 - val_acc: 0.9751\nEpoch 99\/100\n375\/375 [==============================] - 9s 23ms\/step - loss: 0.0643 - acc: 0.9797 - val_loss: 0.0888 - val_acc: 0.9727\nEpoch 100\/100\n375\/375 [==============================] - 9s 23ms\/step - loss: 0.0642 - acc: 0.9796 - val_loss: 0.0939 - val_acc: 0.9707<\/code><\/pre><\/div>\n\n\n\n\u6b63\u89e3\u7387\u3068\u8aa4\u5dee\u95a2\u6570\u306e\u5024\u306e\u5909\u5316\u306e\u7d50\u679c\u3092\u793a\u3057\u307e\u3059\u3002<\/p>\n\n\n
\n
![\"\"](\"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/08\/GRU_results.png\")
<\/figure><\/div>\n\n\n\u7d50\u679c\u304b\u3089\u308f\u304b\u308b\u3053\u3068<\/h3>\n\n\n\n
\u307e\u305a\u306f\u3001\u4e8c\u3064\u306e\u30b0\u30e9\u30d5\u3092\u5408\u308f\u305b\u3066\u307f\u307e\u3057\u3087\u3046\u3002\u4eca\u307e\u3067\u306e\u30b0\u30e9\u30d5\u3060\u3068\u5c11\u3057\u5c0f\u3055\u3044\u306e\u3067\u3001\u305d\u308c\u305e\u308c\u5927\u304d\u304f\u3057\u307e\u3057\u305f\u3002<\/p>\n\n\n
\n
![\"\"](\"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/08\/1dCNN_GRU_results_1.png\")
<\/figure><\/div>\n\n\n\u6b63\u89e3\u7387\u306f\u5168\u4f53\u7684\u306b1\u6b21\u5143CNN\u306e\u65b9\u304c\u9ad8\u3044\u72b6\u614b\u3092\u7dad\u6301\u3057\u3066\u3044\u307e\u3057\u305f\u3002100\u30a8\u30dd\u30c3\u30af\u76ee\u3067\u306f\u4e21\u8005\u306e\u6b63\u89e3\u7387\u306e\u5e45\u304c\u7e2e\u307e\u308a\u307e\u3057\u305f\u304c\u3001\u305d\u308c\u3067\u3082\u30011\u6b21\u5143CNN\u306b\u308f\u305a\u304b\u306b\u8ca0\u3051\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n
\n
![\"\"](\"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/08\/1dCNN_GRU_results_2.png\")
<\/figure><\/div>\n\n\n\u8aa4\u5dee\u95a2\u6570\u306e\u30b0\u30e9\u30d5\u306f\u6b63\u89e3\u7387\u306e\u63a8\u79fb\u3092\u9006\u306b\u3057\u305f\u3088\u3046\u306a\u3082\u306e\u306b\u306a\u308a\u307e\u3057\u305f\u3002<\/p>\n\n\n\n
\u4ed6\u306b\u306f\u3001\u5148\u307b\u3069\u306e\u5b9f\u884c\u7d50\u679c\u306e\u30ed\u30b0\u304b\u3089\u30011\u30a8\u30dd\u30c3\u30af\u306b\u8981\u3059\u308b\u6642\u9593\u306f1\u6b21\u5143CNN\u304c\u7d043\u79d2\u3001GRU\u304c\u7d049\u79d2\u3068\u306a\u3063\u3066\u304a\u308a\u3001\u51e6\u7406\u901f\u5ea6\u306b\u95a2\u3057\u3066\u30821\u6b21\u5143CNN\u306e\u5727\u52dd\u3068\u306a\u308a\u307e\u3057\u305f\u3002<\/p>\n\n\n\n
\u307e\u3068\u3081<\/h2>\n\n\n\n
\u793a\u3057\u305f\u7d50\u679c\u306e\u307f\u304b\u3089\u306f\u3001\u518d\u5e30\u578b\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u3088\u308a1\u6b21\u5143CNN\u306e\u65b9\u304c\u3001\u51e6\u7406\u6642\u9593\u304a\u3088\u3073\u6b63\u89e3\u7387\u306e\u89b3\u70b9\u304b\u3089\u5727\u5012\u7684\u306b\u9ad8\u6027\u80fd\u3067\u3042\u308b\u3053\u3068\u304c\u5206\u304b\u308a\u307e\u3059\u3002\u3068\u306f\u3044\u3048\u30011\u6b21\u5143CNN\u306e\u5834\u5408\u306f\u3001\u8003\u616e\u3067\u304d\u308b\u6642\u523b\u6570\u304c\u30a6\u30a3\u30f3\u30c9\u30a6\u30b5\u30a4\u30ba\u3067\u56fa\u5b9a\u3055\u308c\u308b\u305f\u3081\u3001\u3088\u308a\u9577\u671f\u9593\u306e\u8a18\u61b6\u3092\u4fdd\u6301\u3067\u304d\u308bLSTM\u3084GRU\u306b\u306f\u52a3\u308b\u9762\u304c\u3042\u308b\u3068\u8003\u3048\u308b\u3079\u304d\u3067\u3059\u3002\u305d\u306e\u305f\u3081\u3001\u5165\u529b\u306b\u8fd1\u3044\u5074\u30671\u6b21\u5143CNN\u3092\u3001\u9ad8\u6b21\u306a\u9818\u57df\u3067GRU\u3092\u9069\u7528\u3059\u308b\u306a\u3069\u30012\u3064\u3092\u6df7\u305c\u3066\u6df1\u5c64\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u3092\u69cb\u7bc9\u3059\u308b\u5c0f\u5f0f\u3092\u63a1\u7528\u3059\u308b\u3068\u826f\u3044\u3068\u8003\u3048\u3089\u308c\u307e\u3059\u3002\u6a5f\u4f1a\u304c\u3042\u3063\u305f\u30891\u6b21\u5143CNN+GRU\u306b\u3064\u3044\u3066\u6271\u3048\u308c\u3070\u3068\u601d\u3044\u307e\u3059\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
\u3053\u308c\u3060\u3051\u306f\uff08\u30dd\u30a4\u30f3\u30c8\uff09 \u672c\u8a18\u4e8b\u3067\u8a71\u3059\u5185\u5bb9\u306e\u4e2d\u3067\u3001\u91cd\u8981\u306a\u90e8\u5206\u3092\u30d4\u30c3\u30af\u30a2\u30c3\u30d7\u3057\u3066\u7b87\u6761\u66f8\u304d\u306b\u3057\u307e\u3057\u305f\u3002 \u9806\u4f1d\u64ad\u578b\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\uff08FNN\uff09 \u9806\u4f1d\u64ad\u578b\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\uff08Feedforward Ne … <\/p>\n","protected":false},"author":1,"featured_media":2035,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_vk_print_noindex":"","sitemap_hide":"","_veu_custom_css":"","veu_display_promotion_alert":"common","vkexunit_cta_each_option":"","_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[170,171,106,174],"tags":[154],"class_list":["post-1996","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog","category-learn","category-ml","category-dl","tag-1cnn"],"veu_head_title_object":{"title":"","add_site_title":""},"jetpack_featured_media_url":"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/08\/image.png","jetpack-related-posts":[],"jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/posts\/1996","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/comments?post=1996"}],"version-history":[{"count":72,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/posts\/1996\/revisions"}],"predecessor-version":[{"id":10557,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/posts\/1996\/revisions\/10557"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/media\/2035"}],"wp:attachment":[{"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/media?parent=1996"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/categories?post=1996"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/developers.agirobots.com\/jp\/wp-json\/wp\/v2\/tags?post=1996"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
![\"\"](\"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/07\/image-7-1024x290.png\")
<\/figure><\/div>\n\n\n![\"\"](\"https:\/\/developers.agirobots.com\/jp\/wp-content\/uploads\/2020\/07\/image-7-150x150.png\")
<\/dt>